A distribution scheduling strategy method and system based on artificial bee colony algorithm

Abstract

A distribution scheduling strategy method and system based on an artificial bee colony algorithm, specifically comprising the following steps: S1: active power prediction of demand distribution nodes; S2: active power output and volatility prediction of distributed photovoltaic energy nodes; S3: grid topology model construction; and S4: artificial bee colony distribution scheduling of distributed photovoltaic energy nodes. The present invention proposes a distribution scheduling strategy method based on an artificial bee colony algorithm, which sets a bee colony of appropriate size and performs iterative search to randomly generate active power values ​​at the centralized power generation end and reactive power values ​​in the power grid. By optimizing active power and reactive power, the bee colony fitness value is maximized, thereby selecting the optimal distributed photovoltaic energy node, realizing intelligent control of photovoltaic loads, dynamically adjusting the matching relationship between photovoltaic power generation and loads, achieving regional autonomous consumption, and improving the utilization efficiency of photovoltaic power generation, significantly reducing overall circuit losses and voltage fluctuations, and improving the stability and efficiency of the power system.

Description

Technical Field

[0001] The present invention belongs to the field of power distribution scheduling strategies, and specifically relates to a power distribution scheduling strategy method and system based on an artificial bee colony algorithm. Background Art

[0002] With the growing demand for electricity and the diversification of energy sources in modern society, distributed energy, particularly photovoltaic power generation, is increasingly being used in power systems. As a clean, renewable energy source, distributed photovoltaic power generation has attracted widespread attention and application due to its flexible installation, environmental friendliness, and reduced transmission losses. However, the intermittent and volatile nature of distributed photovoltaic power generation poses new challenges to grid stability and scheduling.

[0003] In traditional power systems, distribution networks rely primarily on centralized power plants to provide a stable power supply. While this centralized generation method ensures a continuous power supply, it carries the risk of single-point failures and significant transmission losses. Centralized power plants are often located far from load centers, requiring long-distance power transmission, resulting in increased transmission losses. Furthermore, centralized power generation relies on fossil fuels, generating significant amounts of carbon dioxide emissions and negatively impacting environmental protection.

[0004] The introduction of distributed photovoltaic power generation can alleviate the pressure on centralized power generation, reduce transmission losses, and minimize environmental pollution. However, distributed photovoltaic power generation is significantly affected by factors such as weather and sunshine, and its output power is highly intermittent and volatile, posing new challenges to grid stability and scheduling.

[0005] Existing power dispatching methods usually rely on forecasting and dispatching models for centralized power generation, which lack effective response strategies for the volatility of distributed photovoltaic power generation. Specifically, traditional forecasting models are often unable to accurately capture the volatility of distributed photovoltaic power generation, resulting in large forecast errors and affecting the accuracy of dispatching decisions. These models are usually based on historical data and statistical methods, and cannot fully consider dynamic factors such as weather changes and sunshine intensity. They also have a slow response speed to real-time changes, making it difficult to adjust power distribution strategies in a timely manner to cope with the rapid changes in distributed photovoltaic power generation. Traditional dispatching systems usually adopt a centralized control method, and the information transmission and decision-making process are relatively cumbersome, and cannot quickly respond to the dynamic changes of distributed photovoltaic power generation;

[0006] Therefore, there is an urgent need for an optimization method that can comprehensively utilize the real-time data of distributed photovoltaic power generation to improve the stability and efficiency of the power system. Summary of the Invention

[0007] To address these issues, the present invention proposes a distribution scheduling strategy method and system based on an artificial bee colony algorithm. By setting a swarm of appropriate sizes and performing an iterative search, the algorithm randomly generates active power values ​​for distributed photovoltaic generation nodes and other distributed energy resources, as well as reactive power values ​​in the power grid, maximizing the swarm's fitness. This method then selects the optimal distributed energy resource nodes, reducing overall circuit losses and voltage fluctuations. This method effectively utilizes real-time data from distributed photovoltaic generation, improving the stability and efficiency of the power system.

[0008] To achieve the above object, the technical solution adopted by the present invention is:

[0009] A distribution scheduling strategy method based on artificial bee colony algorithm, the specific steps are as follows:

[0010] S1: Active power forecast of demand distribution nodes;

[0011] To predict the active power value and trend of the demand distribution node, it is necessary to consider the timely and transient access of distributed energy resources. The active power value and trend of the demand distribution node are predicted and fed back. That is, the active power value and trend of the demand distribution node within one minute are predicted. The active power trend value of the demand distribution node is the active power of the demand distribution node 30 seconds ago minus the active power of the current demand distribution node. Separate models are set for different regions, and normalized general characteristics are not considered.

[0012] The smart meter is used to read the active power of the real-time power consumption data of the demand distribution node every second, in kilowatt-hours, and recorded as W p , read the weather conditions of the nearby weather station, including the real-time temperature T, humidity S, and wind speed F of the area, and use one-hot encoding to record the time scale as a D vector. This vector includes the hour, minute, day of the week, and whether it is a holiday. The scale value that the specific minute is closest to is used as the mark value, the day of the week is marked as Monday to Sunday, and whether it is a holiday is marked as yes or no;

[0013] S2: Active power output and fluctuation prediction of distributed photovoltaic energy nodes;

[0014] Step S2 predicts the active power output and fluctuation of distributed photovoltaic energy nodes;

[0015] First, a distributed solar photovoltaic prediction time series model near the demand distribution node is established to predict the distributed solar photovoltaic active power change trend value and the specific solar photovoltaic active power value. The distributed solar photovoltaic active power change trend value is the difference between the real-time solar photovoltaic active power and the distributed solar photovoltaic active power 30 seconds ago.

[0016] Among them, the distributed solar photovoltaic prediction timing model is expressed as follows:

[0017] Solar tend ,Solar energy =Rnn(Soir,T,S,F,G,Clc)

[0018] Among them, Solar tend Indicates the trend value of solar photovoltaic active power change, Solar energy represents the solar energy in the next second, Rnn represents the recurrent neural network, Soir represents the real-time sunshine intensity, Clc represents the real-time cloud cover, G represents the solar radiation intensity, and Clc represents the real-time cloud cover;

[0019] S3: Grid topology model construction;

[0020] Determine the location of the demand distribution node loc x , and confirm the location of each distributed energy source loc f , and then the impedance R of the power grid is approximately calculated by the impedance calculation formula i :

[0021] R i =R line *L+j*X line *L

[0022] Among them, R i is the grid impedance, R line is the line resistance, j is the imaginary unit, X line is the line reactance, L is the line length from the distributed energy node to the demand distribution node;

[0023] S4: artificial swarm power distribution scheduling of distributed energy nodes;

[0024] A power dispatching method for distributed energy nodes based on artificial bee colony algorithm is proposed.

[0025] Furthermore, the active power prediction process of the demand distribution node in step S1 includes enhancing the dynamic load factor formula, integrating time series characteristics, and considering the active power usage fluctuation of the demand distribution node in a short period of time:

[0026]

[0027] Among them, DIF enhaced To enhance the dynamic load factor, p avg Ask the average active power in the last 30 seconds, p maxis the maximum active power within 30 seconds in real time, t is the time of the day in real time, in hours, T is the total number of hours in a day, which is 24, ε and δ are weight factors, G is the solar radiation intensity, P pv is the output power of the photovoltaic power generation system, ΔP pv The change in the output power of the photovoltaic power generation system (i.e. the current P pv Subtract P from 30 seconds ago pv ), γ, ρ and θ are the weight factors of photovoltaic power generation;

[0028] Furthermore, the active power prediction process of the demand distribution node in step S1 includes characterizing the instantaneous efficiency of the transformer. When the ratio of the instantaneous output power to the input power of the transformer is low, although there is no direct active power demand at the demand distribution node, the overall energy consumption of the power grid will increase, thereby requiring more active power:

[0029]

[0030] Among them, Tran effic is the instantaneous efficiency of the transformer, p in (t),p out (t) are the input power and output power at time t, p' in (t) and p' out (t) is the instantaneous rate of change of input power and output power, where the sampling time interval of the rate of change is 30 seconds.

[0031] Furthermore, in the step S1, the Transformer network is used to predict the real-time active power value and trend of the demand distribution node in the next second, wherein the Transformer network is represented as:

[0032] Acticep tend , Acticep z =Transformer(W p ,T,S,F,D,DIF enhaced ,P pv ,ΔP pv ,Tran effic )

[0033] Among them, Acticep tend Indicates active power trend, Acticep z Indicates the active power value of the demand distribution node in the next second. Transformer is a sequential network architecture. pv is the output power of the photovoltaic power generation system, ΔP pv is the change in output power of the photovoltaic power generation system;

[0034] Furthermore, in the step S2, a wind energy prediction time series model is established for the vicinity of the demand distribution node during the distributed energy node active power output and volatility prediction process to predict the wind energy active power change trend value and the specific wind energy active power value, wherein the wind energy active power change trend value is specifically expressed as the difference between the wind energy active power value and the wind energy active power value 30 seconds ago;

[0035] Wind tend ,Wind energy =Rnn(F,T,Ap,Wd)

[0036] Among them, Wind tend is the trend value of wind energy active power change, Wind energy is the active power value of wind energy in the next second, Ap is the air pressure, and Wd is the wind direction.

[0037] Furthermore, the specific steps of the S4 distributed energy node artificial bee colony power distribution scheduling are as follows:

[0038] Step A1: Distributed energy fluctuation threshold processing;

[0039] In this step, a distributed energy fluctuation threshold function is first proposed to consider all distributed energy resources near the demand distribution node. If the volatility is too high, scheduling is directly abandoned:

[0040]

[0041] Among them, C i is the judgment value of the i-th distributed energy node. When the value is 0, its supply right is cancelled, and when it is 1, its supply is considered. Cv i is the active power change trend value of the i-th distributed energy node, Cv max is the upper limit of the active power change trend of the distributed energy node;

[0042] Step A2: Artificial bee colony algorithm power optimization;

[0043] In this step, we first determine an artificial bee colony algorithm power optimization formulation and propose a dynamic matching response function:

[0044]

[0045] Among them, Fle i is the dynamic demand matching value of the i-th distributed energy node, Ct i is the real-time active power of the i-th distributed energy node;

[0046] Based on the dynamic matching response function above, the distributed energy nodes with negative dynamic demand matching values ​​are directly used as backup energy supply nodes, and the distributed energy nodes with positive dynamic demand matching values ​​close to 0 are considered as energy response nodes;

[0047] Step A3: Artificial bee colony algorithm reactive power setting;

[0048] In step A2 above, the specific supply points of the distributed energy nodes are determined. Since the artificial bee colony algorithm is used to optimize the active power of the centralized power generation end and the reactive power in the power grid, the voltage is optimized and the overall circuit power loss is reduced;

[0049] In the artificial bee colony algorithm, the solution includes the active power of the centralized power generation terminal and the reactive power value in the power grid:

[0050] x=(p centeral ,Q grid )

[0051] Where x is the solution of reactive power optimization by artificial bee colony algorithm, p centeral is the active power of the centralized power generation terminal, Q grid is the reactive power value in the power grid;

[0052] According to the solution of reactive power optimization of artificial bee colony algorithm, the voltage of demand distribution node is approximately calculated:

[0053]

[0054] Among them, V t is the adjusted voltage of the demand distribution node, and I is the demand distribution current;

[0055] Based on this, in this step, a fitness function is proposed to give the target value of the artificial bee colony algorithm:

[0056]

[0057] Among them, Fit is the fitness of the bee colony, R i is the grid impedance, I is the current value, V t is the voltage after adjustment of the demand distribution node, V ideal The ideal voltage of the demand distribution node;

[0058] Step A4: The artificial bee colony algorithm finally finds the optimal solution.

[0059] A power distribution scheduling strategy system based on artificial bee colony algorithm, including the following modules:

[0060] Data acquisition module: The data acquisition module is used to install sensors in the distributed photovoltaic power generation system to collect the output power of the photovoltaic power generation system in real time, including the current, voltage and power data of the photovoltaic panels. At the same time, it obtains real-time weather data through the meteorological station, including temperature, humidity, wind speed and solar radiation intensity. This data is used to analyze and predict the output of photovoltaic power generation and is connected with other modules to ensure the accuracy and real-time nature of the data.

[0061] Data preprocessing module: The data preprocessing module is used to process the collected photovoltaic power output power data and weather data, normalize the power data and calculate the power change trend, denoise, standardize and extract features of the weather data, and generate environmental feature data for model training to ensure data consistency and comparability for subsequent model use;

[0062] Prediction model module: The prediction model module includes enhanced dynamic load factor calculation and Transformer prediction model. By calculating the enhanced dynamic load factor, it comprehensively considers the real-time fluctuations of photovoltaic power generation, weather factors and changes in power demand to generate features for load forecasting. Using the Transformer network model, the module predicts the active power value and trend of the demand distribution node in real time. The input data includes pre-processed power data, environmental data and enhanced dynamic load factor.

[0063] Grid topology module: The grid topology module is used to determine the locations of demand distribution nodes, distributed photovoltaic power generation nodes, and other distributed energy nodes, establish a grid topology model, and accurately simulate the operation status of the grid by comprehensively considering the impact of photovoltaic power generation nodes and other distributed energy nodes through impedance calculation;

[0064] Distribution Scheduling Module: This module is used to assess the power fluctuations of distributed photovoltaic power generation nodes. When the fluctuations are too large, their supply qualifications are revoked. An artificial bee colony algorithm is used to optimize power. The optimal distributed energy node is determined by dynamically matching the response function to improve the efficiency of the power system. At the same time, the reactive power in the power grid is optimized, and the active power of the centralized power generation end and the distributed photovoltaic power generation nodes is combined to reduce grid losses and voltage fluctuations.

[0065] Feedback and Optimization Module: The feedback and optimization module feeds prediction results and scheduling plans back to the control system in real time, dynamically adjusts the power distribution strategy, ensures the stability and reliability of power supply, and continuously optimizes the artificial bee colony algorithm and scheduling strategy based on feedback data to improve the overall efficiency and stability of the system and ensure the optimized operation of the power system.

[0066] Beneficial effects:

[0067] (1) This solution can more accurately capture the volatility of photovoltaic power generation by collecting and processing the output power and related environmental data of distributed photovoltaic power generation systems in real time, enhance the application of dynamic load factors and Transformer network models, and significantly improve the prediction accuracy of the active power value and trend of demand distribution nodes. The artificial bee colony algorithm is used for distribution scheduling, which can quickly respond to real-time changes in distributed photovoltaic power generation, dynamically adjust the distribution strategy, and ensure the stability and continuity of power supply.

[0068] (2) This solution effectively improves the overall scheduling optimization effect by comprehensively considering the output data of different types of distributed energy and utilizing multimodal data fusion technology. The application of the artificial bee colony algorithm enables the system to find the optimal distributed energy node in a dynamic environment, reducing the overall loss and voltage fluctuation of the power grid and improving the operating efficiency of the power grid. The introduction of distributed photovoltaic power generation systems reduces the long-distance transmission loss of electricity and further improves energy utilization.

[0069] (3) This solution reduces power waste and equipment loss caused by improper scheduling through accurate load forecasting and optimized distribution scheduling strategies. The effective utilization of distributed photovoltaic power generation systems reduces dependence on centralized power generation, reduces power generation and transmission costs, and thus reduces the overall operating costs of the power system. This not only improves economic benefits but also conforms to the development concept of green environmental protection. BRIEF DESCRIPTION OF THE DRAWINGS

[0070] Figure 1 It is the overall flow chart of the algorithm;

[0071] Figure 2 This is the algorithm power optimization flow chart of this method. DETAILED DESCRIPTION

[0072] The present invention is further described in detail below with reference to the accompanying drawings and specific embodiments:

[0073] The present invention proposes a flow chart such as Figure 1 The algorithm power optimization flow chart is as follows Figure 2 A distribution scheduling strategy method and system based on artificial bee colony algorithm is shown, and the specific implementation method is as follows;

[0074] The specific implementation is as follows;

[0075] S1: Active power forecast of demand distribution nodes;

[0076] In this step, the active power value and trend of the demand distribution node are predicted, but the difference is that this application needs to consider the timely transient distributed photovoltaic energy access, so this application needs to be able to predict and feedback the active power value and trend of the demand distribution node in real time and quickly, that is, to predict the active power value and trend of the demand distribution node within minutes, where the active power trend of the demand distribution node is the active power of the demand distribution node 30 seconds ago minus the active power of the current demand distribution node. Therefore, this application sets up separate models for different areas for consideration, and does not consider normalized general characteristics, such as the population size of the area, which will not affect the demand distribution area within minutes;

[0077] Therefore, based on the above application, it is proposed to use smart meters to read the active power of the real-time power consumption data of the demand distribution node every second, in kilowatt-hours, and record it as W p , read the weather conditions of the nearby weather station, including the real-time temperature T, humidity S, wind speed F and solar radiation intensity G in the area, and record the output power of the nearby distributed photovoltaic power generation system in kilowatt-hours, and record it as P pv In addition, the time scale at this time is recorded as a D vector using one-hot encoding. This vector includes hours, minutes, day of the week, and whether it is a holiday. Specifically, the hours are marked as 0-23, and the minutes are 00, 15, 30, and 45. The scale value that the specific minute is closest to is used as the mark value. The day of the week is marked as Monday to Sunday, and whether it is a holiday is marked as yes or no.

[0078] Calculate the output power trend of the photovoltaic power generation system, that is, P 30 seconds ago pv Subtract the current P pv value.

[0079] Furthermore, this application proposes an enhanced dynamic load factor formula that integrates time series characteristics and considers the fluctuation of active power usage of demand distribution nodes within a short period of time:

[0080]

[0081] Among them, DIF enhaced To enhance the dynamic load factor, p avg Ask the average active power in the last 30 seconds, p max is the maximum active power within 30 seconds in real time, t is the time of the day in real time, in hours, T is the total number of hours in a day, which is 24, ε and δ are weight factors, G is the solar radiation intensity, P pv is the output power of the photovoltaic power generation system, ΔP pv The change in the output power of the photovoltaic power generation system (i.e. the current Ppv Subtract P from 30 seconds ago pv ), γ, ρ and θ are the weight factors of photovoltaic power generation;

[0082] Furthermore, for example, through statistical analysis, ε and δ in a residential area were normalized at each time node every day in a certain summer, and linear regression was performed on all active power values, and it was obtained that ε was 0.3 and δ was 0.5, that is, the overall time was at noon and evening, and the active power demand was high at this time, which amplified the active power fluctuation value of the demand distribution node.

[0083] Furthermore, this application characterizes the instantaneous efficiency of the transformer. When the ratio of the transformer's instantaneous output power to input power is low, although there is no direct active power demand at the demand distribution node, the overall energy consumption of the grid will increase, and more active power will be required:

[0084]

[0085] Among them, Tran effic is the instantaneous efficiency of the transformer, p in (t),p out (t) are the input power and output power at time t, p' in (t) and p' out (r) is the instantaneous rate of change of input power and output power, where the sampling interval of the rate of change is 30 seconds;

[0086] Based on the above, this application uses the Transformer network to predict the real-time active power value and its trend of the demand distribution node in the next second. The Transformer network is represented as:

[0087] Acticep tend , Acticep z =Transformer(W p ,T,S,F,D,DIF enhaced ,P pv ,ΔP pv ,Tran effic )

[0088] Among them, Acticep tend Indicates active power trend, Acticep z Indicates the active power value of the demand distribution node in the next second. Transformer is a sequential network architecture. pv is the output power of the photovoltaic power generation system, ΔP pv is the change in output power of the photovoltaic power generation system;

[0089] S2: Active power output and fluctuation prediction of distributed photovoltaic energy nodes;

[0090] In step S1, this application predicts the active power and its trend of the demand distribution node. Further, in this step, it is necessary to predict the active power and volatility of the distributed photovoltaic power generation nodes and other distributed energy nodes;

[0091] There are usually many distributed energy nodes near the demand distribution nodes, including solar photovoltaic power generation, wind power generation, small hydropower generation, etc. Among them, solar photovoltaic power generation is highly intermittent and unstable due to changes in sunshine intensity and weather, while changes in wind speed in wind power generation will also cause large fluctuations in its active power provision capacity.

[0092] Based on the above, in this step, a distributed solar photovoltaic prediction time series model is first established near the demand distribution node to predict the distributed solar photovoltaic active power change trend value and the specific solar photovoltaic active power value. The distributed solar photovoltaic active power change trend value is the difference between the real-time solar photovoltaic active power and the distributed solar photovoltaic active power 30 seconds ago.

[0093] Among them, the distributed solar photovoltaic prediction timing model is expressed as follows:

[0094] Solar tend ,Solar energy =Rnn(Soir,T,S,F,G,Clc)

[0095] Among them, Solar tend Indicates the trend value of solar photovoltaic active power change, Solar energy represents the solar energy in the next second, Rnn represents the recurrent neural network, Soir represents the real-time sunshine intensity, Clc represents the real-time cloud cover, and G represents the solar radiation intensity;

[0096] Furthermore, a wind energy prediction timing model near the demand distribution node is established to predict the active power change trend value of wind energy and the specific active power value of wind energy. The active power change trend value of wind energy can be specifically expressed as the difference between the wind energy active power value and the wind energy active power 30 seconds ago.

[0097] Wind tend ,Wind energy =Rnn(F,T,Ap,Wd)

[0098] Among them, Wind tend is the trend value of wind energy active power change, Wind energyis the active power value of wind energy in the next second, Ap is the air pressure, and Wd is the wind direction;

[0099] S3: Grid topology model construction;

[0100] Determine the location of the demand distribution node loc x , and confirm the location of each distributed photovoltaic power generation node and other distributed energy nodes f , and then the impedance R of the power grid is approximately calculated by the impedance calculation formula i :

[0101] R i =R line *L+j*X line *L

[0102] Among them, R i is the grid impedance, R line is the line resistance, j is the imaginary unit, X line is the line reactance, L is the line length from the distributed energy node to the demand distribution node;

[0103] S4: artificial swarm power distribution scheduling of distributed photovoltaic energy nodes;

[0104] In step S1, this application predicts the active power value and its trend of the demand distribution node, and in step S2, completes the active power change trend value and specific active power value generated by the distributed photovoltaic power generation node and other distributed energy nodes. Based on this, this step proposes power scheduling of distributed energy nodes based on the artificial bee colony algorithm;

[0105] Wherein, step S4 includes the following sub-steps:

[0106] Step A1: Distributed energy fluctuation threshold processing;

[0107] In this step, a distributed energy fluctuation threshold function is first proposed to consider all distributed photovoltaic power generation nodes and other distributed energy nodes near the demand distribution node. If the fluctuation is too high, scheduling is directly abandoned:

[0108]

[0109] Among them, C i is the judgment value of the i-th distributed energy node. When the value is 0, its supply right is cancelled, and when it is 1, its supply is considered. Cv i is the active power change trend value of the i-th distributed energy node, Cv max It is the upper limit of the active power change trend of the distributed energy node.

[0110] Step A2: Artificial bee colony algorithm power optimization

[0111] In this step, we first determine an artificial bee colony algorithm power optimization formulation and propose a dynamic matching response function:

[0112]

[0113] Among them, Fle i is the dynamic demand matching value of the i-th distributed energy node, Ct i is the real-time active power of the i-th distributed energy node;

[0114] Based on the dynamic matching response function above, the distributed energy nodes with negative dynamic demand matching values ​​are directly used as backup energy supply nodes, and the distributed energy nodes with positive dynamic demand matching values ​​close to 0 are considered as energy response nodes.

[0115] Step A3: Artificial bee colony algorithm reactive power setting

[0116] In step A2 above, this application determines the specific supply points of distributed energy nodes. Based on this, this application uses an artificial bee colony algorithm to optimize the active power of the centralized power generation end and the reactive power in the power grid, optimize the voltage and reduce the overall circuit power loss;

[0117] In the artificial bee colony algorithm, the solution includes the active power of the centralized power generation terminal and the reactive power value in the power grid:

[0118] x=(p centeral ,Q grid )

[0119] Where x is the solution of reactive power optimization by artificial bee colony algorithm, p centeral is the active power of the centralized power generation terminal, Q grid is the reactive power value in the power grid;

[0120] Furthermore, the voltage of the demand distribution node is approximately calculated based on the reactive power optimization solution of the artificial bee colony algorithm:

[0121]

[0122] Among them, V t is the adjusted voltage of the demand distribution node, and I is the demand distribution current;

[0123] Based on this, in this step, a fitness function is proposed to give the target value of the artificial bee colony algorithm:

[0124]

[0125] Among them, Fit is the fitness of the bee colony, R i is the grid impedance, I is the current value, V t is the voltage after adjustment of the demand distribution node, V ideal The ideal voltage of the demand distribution node;

[0126] Step A4: The artificial bee colony algorithm finally finds the optimal solution.

[0127] Based on the above, this application proposes a swarm size of 30 and performs an iterative search to randomly generate active power at the generator-type power generation end and reactive power values ​​in the power grid to maximize the swarm fitness value, thereby selecting the optimal distributed energy node and reducing overall circuit losses and voltage fluctuations.

[0128] The above description is merely a preferred embodiment of the present invention and does not constitute any other form of limitation to the present invention. Any modification or equivalent variation based on the technical essence of the present invention shall still fall within the scope of protection claimed by the present invention.

Claims

1. A power distribution scheduling strategy method based on artificial bee colony algorithm, characterized in that: The specific steps are as follows: S1: Active power forecast of demand distribution nodes; To predict the active power value and trend of the demand distribution node, it is necessary to consider the timely and transient access of distributed energy resources. The active power value and trend of the demand distribution node are predicted and fed back. That is, the active power value and trend of the demand distribution node within one minute are predicted. The active power trend of the demand distribution node is the active power of the demand distribution node 30 seconds ago minus the active power of the current demand distribution node. Separate models are set for different regions to take into account, and normalized general characteristics are not considered. The smart meter is used to read the active power of the real-time power consumption data of the demand distribution node every second, in kilowatt-hours, and recorded as W p , read the weather conditions of the nearby weather station, including the real-time temperature T, humidity S, and wind speed F of the area, and use one-hot encoding to record the time scale as a D vector. This vector includes the hour, minute, day of the week, and whether it is a holiday. The scale value that the specific minute is closest to is used as the mark value, the day of the week is marked as Monday to Sunday, and whether it is a holiday is marked as yes or no; In the step S1, during the active power prediction process of the demand distribution node, the Transformer network is used to predict the real-time active power value and trend of the demand distribution node in the next second; S2: Active power output and fluctuation prediction of distributed photovoltaic energy nodes; Step S2 predicts the active power output and fluctuation of distributed photovoltaic energy nodes; First, a distributed solar photovoltaic prediction time series model near the demand distribution node is established to predict the distributed solar photovoltaic active power change trend value and the specific solar photovoltaic active power value. The distributed solar photovoltaic active power change trend value is the difference between the real-time solar photovoltaic active power and the distributed solar photovoltaic active power 30 seconds ago. Among them, the distributed solar photovoltaic prediction timing model is expressed as follows: Supply tend ,Supply energy =Rnn(Soir,T,S,F,G,Clc) Among them, Solar tend Indicates the trend value of solar photovoltaic active power change, Solar energy represents the solar energy in the next second, Rnn represents the recurrent neural network, Soir represents the real-time sunshine intensity, Clc represents the real-time cloud cover, and G represents the solar radiation intensity; S3: Grid topology model construction; Determine the location of the demand distribution node loc x , and confirm the location of each distributed energy source loc f , and then the impedance R of the power grid is approximately calculated by the impedance calculation formula i : R i =R line *L+j*X line *L Among them, R i is the grid impedance, R line is the line resistance, j is the imaginary unit, X line is the line reactance, L is the line length from the distributed energy node to the demand distribution node; S4: artificial swarm power distribution scheduling of distributed energy nodes; A power dispatching method for distributed energy nodes based on artificial bee colony algorithm is proposed.

2. The power distribution scheduling strategy method based on artificial bee colony algorithm according to claim 1 is characterized in that: The active power prediction process of the demand distribution node in step S1 includes enhancing the dynamic load factor formula, integrating time series characteristics, and considering the active power usage fluctuation of the demand distribution node within a short period of time: Among them, DIF enhaced To enhance the dynamic load factor, p avg is the average active power within 30 seconds in real time, p max is the maximum active power within 30 seconds in real time, t is the time of day in real time, in hours, T′ is the total number of hours in a day, which is 24, ε and δ are weight factors, G is the solar radiation intensity, P pv is the output power of the photovoltaic power generation system, ΔP pv is the change in output power of the photovoltaic power generation system, that is, the current P pv Subtract P from 30 seconds ago pv , γ, ρ and θ are the weight factors of photovoltaic power generation.

3. The power distribution scheduling strategy method based on artificial bee colony algorithm according to claim 2 is characterized in that: The active power prediction process of the demand distribution node in step S1 includes characterizing the instantaneous efficiency of the transformer. When the ratio of the instantaneous output power to the input power of the transformer is low, although there is no direct active power demand at the demand distribution node, the overall energy consumption of the power grid will increase, thereby requiring more active power: Among them, Tran effic is the instantaneous efficiency of the transformer, p in (t),p out (t) are the input power and output power at time t, p' in (t) and p' out (t) is the instantaneous rate of change of input power and output power, where the sampling time interval of the rate of change is 30 seconds.

4. The power distribution scheduling strategy method based on artificial bee colony algorithm according to claim 3 is characterized in that: The Transformer network in step S1 is represented as: Acticep tend ,Acticep z =Transformer(W p ,T,S,F,D,DIF enhaced ,P pv ,ΔP pv ,Tran effic ) Among them, Acticep tend Indicates active power trend, Acticep z Indicates the active power value of the demand distribution node in the next second. Transformer is a temporal network architecture. pv is the output power of the photovoltaic power generation system, ΔP pv It is the change in output power of the photovoltaic power generation system.

5. The power distribution scheduling strategy method based on artificial bee colony algorithm according to claim 4 is characterized in that: In the step S2, a wind energy prediction time series model is established for the distributed energy node active power output and volatility prediction process to predict the wind energy active power change trend value and the specific active power value of the wind energy, wherein the wind energy active power change trend value is specifically expressed as the difference between the wind energy active power value and the wind energy active power value 30 seconds ago; Wind tend ,Wind energy =Rnn(F,T,Ap,Wd) Among them, Wind tend is the trend value of wind energy active power change, Wind energy is the active power value of wind energy in the next second, Ap is the air pressure, and Wd is the wind direction.

6. The power distribution scheduling strategy method based on artificial bee colony algorithm according to claim 5 is characterized in that: The specific steps of the S4 distributed energy node artificial bee colony power distribution scheduling are as follows: Step A1: Distributed energy fluctuation threshold processing; In this step, a distributed energy fluctuation threshold function is first proposed to consider all distributed energy resources near the demand distribution node. If the volatility is too high, scheduling is directly abandoned: Among them, C i is the judgment value of the i-th distributed energy node. When the value is 0, its supply right is cancelled, and when it is 1, its supply is considered. Cv i is the active power change trend value of the i-th distributed energy node, Cv max is the upper limit of the active power change trend of the distributed energy node; Step A2: Artificial bee colony algorithm power optimization; In this step, we first determine an artificial bee colony algorithm power optimization strategy and propose a dynamic matching response function: Among them, Fle i is the dynamic demand matching value of the i-th distributed energy node, Ct i is the real-time active power of the i-th distributed energy node; Based on the dynamic matching response function above, the distributed energy nodes with negative dynamic demand matching values ​​are directly used as backup energy supply nodes, and the distributed energy nodes with positive dynamic demand matching values ​​that are closest to 0 are considered as energy response nodes; Step A3: Artificial bee colony algorithm reactive power setting; In step A2 above, the specific supply points of the distributed energy nodes are determined. Since the artificial bee colony algorithm is used to optimize the active power of the centralized power generation end and the reactive power in the power grid, the voltage is optimized and the overall circuit power loss is reduced; In the artificial bee colony algorithm, the solution includes the active power of the centralized power generation terminal and the reactive power value in the power grid: x=(p centeral ,Q grid ) Where x is the solution of reactive power optimization by artificial bee colony algorithm, p centeral is the active power of the centralized power generation terminal, Q grid is the reactive power value in the power grid; According to the solution of reactive power optimization of artificial bee colony algorithm, the voltage of demand distribution node is approximately calculated: Among them, V t is the adjusted voltage of the demand distribution node, and I is the demand distribution current; Based on this, in this step, a fitness function is proposed to give the target value of the artificial bee colony algorithm: Among them, Fit is the fitness of the bee colony, R i is the grid impedance, I is the current value, V t is the voltage after adjustment of the demand distribution node, V ideal The ideal voltage for the demand distribution node; Step A4: The artificial bee colony algorithm finally finds the optimal solution.

7. A power distribution scheduling strategy system based on an artificial bee colony algorithm for implementing the method according to any one of claims 1 to 6, characterized in that: Includes the following modules: Data acquisition module: The data acquisition module is used to install sensors in the distributed photovoltaic power generation system to collect the output power of the photovoltaic power generation system in real time, including the current, voltage and power data of the photovoltaic panels. At the same time, it obtains real-time weather data through the meteorological station, including temperature, humidity, wind speed and solar radiation intensity. This data is used to analyze and predict the output of photovoltaic power generation and is connected with other modules to ensure the accuracy and real-time nature of the data. Data preprocessing module: The data preprocessing module is used to process the collected photovoltaic power output power data and weather data, normalize the power data and calculate the power change trend, denoise, standardize and extract features of the weather data, and generate environmental feature data for model training to ensure data consistency and comparability for subsequent model use; Prediction model module: The prediction model module includes enhanced dynamic load factor calculation and Transformer prediction model. By calculating the enhanced dynamic load factor, it comprehensively considers the real-time fluctuations of photovoltaic power generation, weather factors and changes in power demand to generate features for load forecasting. Using the Transformer network model, the module predicts the active power value and trend of the demand distribution node in real time. The input data includes pre-processed power data, environmental data and enhanced dynamic load factor. Grid topology module: The grid topology module is used to determine the locations of demand distribution nodes, distributed photovoltaic power generation nodes, and other distributed energy nodes, establish a grid topology model, and accurately simulate the operation status of the grid by comprehensively considering the impact of photovoltaic power generation nodes and other distributed energy nodes through impedance calculation; Distribution Scheduling Module: This module is used to assess the power fluctuations of distributed photovoltaic power generation nodes. When the fluctuations are too large, their supply qualifications are revoked. An artificial bee colony algorithm is used to optimize power. The optimal distributed energy node is determined by dynamically matching the response function to improve the efficiency of the power system. At the same time, the reactive power in the power grid is optimized, and the active power of the centralized power generation end and the distributed photovoltaic power generation nodes is combined to reduce grid losses and voltage fluctuations. Feedback and Optimization Module: The feedback and optimization module feeds prediction results and scheduling plans back to the control system in real time, dynamically adjusts the power distribution strategy, ensures the stability and reliability of power supply, and continuously optimizes the artificial bee colony algorithm and scheduling strategy based on feedback data to improve the overall efficiency and stability of the system and ensure the optimized operation of the power system.

Images